The specific aims of the project are to develop an optimized technique for quantitative analysis of tumor volume in response to treatment. This will be primarily accomplished using various strategies for supervised and unsupervised segmentation of image data. The data will be generated from registered multi-modality (MRI and PET) images of patients with tumors of the head and neck and brain who are undergoing treatment. This project will produce objective methods to determine tumor volumes which are critically important for therapy planning and monitoring. The result will be increased confidence in the three-dimensional determination of the boundaries of active tumor tissue, edema, and necrosis with a high degree of objectivity.